On 3-Jun-07, at 1:53 AM, Cyril Plisko wrote:
>> That being said, I suppose it could be run in a bazzar fashion ( like
>> Linux ) where there is no such thing as a code lock & there's a race
>> to be first to commit for credit. On the one hand that risks
>> duplication of effort ( and with such a small dev. team that's not
>> really a concern you can just ignore ), but on the other hand we
>> don't get the result we have now ( stalled development because
>> someone's working on code ). If I recall correctly one of the NetBSD
>> founders regrets choosing the second model for precisely the reason I
>> just attempted to outline.
>>
>> I've personally not decided which is the better methodology at the
>> moment... comments?
>
> Which methodologies exactly are you comparing ? It is not clear to me.
> I also think that we can work out the exact methodology as we go. And
> even change it on the fly if we think it serves us better. No need to
> carve it in stone beforehand.

The two methodologies, for lack of a better name are Linux style and  
NetBSD style.

Linux style development, the code doesn't exist until it's been  
committed to the tree, nobody's "working on it" until there's actual  
code there, and the first person to finish it ( where "it" is any  
task that needs to be done ) commits the code. The problem with this  
is you can potentially have multiple people working on a single task,  
only one of whom can finish first, and the rest are just wasting  
their time but you don't know who's in which camp.  On the other  
hand, the bonuses are that code gets done & you don't end up with the  
problem we've had so far ( "Sun Labs is working on a bunch of stuff  
that we don't know what it is... better not risk duplicating the  
code" ). It's quite anarchic, but it seems to work for some projects.

NetBSD style development, someone's assigned a task  ( even if it's  
informally, like "hey guys, i'm working on $foo" ) and it's their job  
to get that finished, or notify everyone that they can't do it. The  
problem with this methodology is that you can end up with code stalls  
like we've had so far, waiting for one member to finish something for  
damn-near forever. On the other hand, you don't end up with two  
people working on the same thing at the same time, one of whom is  
doomed to be wasting their time.

This is more or less a caricature of how Linux & NetBSD /actually/  
work, but I just mean to outline the two development philosophies.

And yeah, it's not set in stone, but I think it does need some  
discussion as to how we're to approach development initially

--
This Message has been Digitally Signed

PGP Public Key 0x437AF1A1
Availiable on hkp://pgp.mit.edu



-------------- next part --------------
A non-text attachment was scrubbed...
Name: PGP.sig
Type: application/pgp-signature
Size: 186 bytes
Desc: This is a digitally signed message part
URL: 
<http://mail.opensolaris.org/pipermail/powerpc-discuss/attachments/20070603/dfe61a9f/attachment.bin>

Reply via email to